Packages

o

scalation.analytics.forecaster

RollingValidation

object RollingValidation

The RollingValidation object provides 'k'-fold rolling validations, e.g., for 'm = 1200' and 'k = 10', 'kt = 20':

1: tr(ain) 0 until 800, te(st) 800 until 840 2: tr(ain) 40 until 840, te(st) 840 until 880 3: tr(ain) 80 until 880, te(st) 880 until 920 4: tr(ain) 120 until 920, te(st) 920 until 960 5: tr(ain) 160 until 960, te(st) 960 until 1000 6: tr(ain) 200 until 1000, te(st) 1000 until 1040 7: tr(ain) 240 until 1040, te(st) 1040 until 1080 8: tr(ain) 280 until 1080, te(st) 1080 until 1120 9: tr(ain) 320 until 1120, te(st) 1120 until 1160 10: tr(ain) 360 until 1160, te(st) 1160 until 1200

In rolling validation for this case, each training dataset has 800 instances, while each testing dataset has 40. Re-training occurs before every 'kt = 20' forecasts are made (2 re-trainings per testing dataset for this case).

Linear Supertypes
AnyRef, Any
Ordering
  1. Alphabetic
  2. By Inheritance
Inherited
  1. RollingValidation
  2. AnyRef
  3. Any
  1. Hide All
  2. Show All
Visibility
  1. Public
  2. All

Value Members

  1. def crossValidate(model: PredictorMat with ForecasterMat, k: Int = 5, kt_: Int = 10, h: Int = 1): Array[Statistic]
  2. def crossValidate2(model: ForecasterVec, k: Int = 5, kt_: Int = 10, h: Int = 1): Array[Statistic]
  3. def crossValidate3(model: PredictorMat2 with ForecasterMat, k: Int = 5, kt_: Int = 10, h: Int = 1): Array[Statistic]
  4. def trSize(m: Int): Int

    Calculate the size (number of instances) for a training dataset.

    Calculate the size (number of instances) for a training dataset.

    m

    the size of the full dataset